How to Sample Without Replacement in R

March 15, 2026 · 3 min read · Updated March 15, 2026 · beginner

r sample random without-replacement dplyr vector

Sampling without replacement means each element can only be selected once. This is the standard approach for creating random subsets, randomized lists, and train/test splits.

The sample() Function

Base R’s sample() does this by default:

# Sample 3 numbers from 1 to 10 without replacement
sample(1:10, size = 3)
# [1] 4 9 2

# Sample 5 letters without replacement
sample(letters, size = 5)
# [1] "u" "q" "z" "k" "m"

# Shuffle all elements (randomize order)
sample(1:10)
# [1]  3  9  5 10  6  1  2  4  8  7

The default behavior is replace = FALSE, meaning no element appears twice.

With dplyr

The tidyverse approach uses slice_sample() on a vector or sample() within mutate():

library(dplyr)

# Using slice_sample on a tibble column
df <- tibble(values = 1:20)
sampled <- df %>% slice_sample(n = 5)

# Sample from a vector directly
sampled <- sample_n(tibble(x = 1:100), 10)
# A tibble: 10 × 1
#       x
#    <int>
# 1    23
# 2    87
# 3    12
# 4    56
# 5    91
# 6    34
# 7     8
# 8    45
# 9    67
# 10   78

Practical Examples

Create a Random Test Set

# Create indices for train/test split
set.seed(42)
n <- 100
indices <- 1:n

train_indices <- sample(indices, size = 70)
test_indices <- setdiff(indices, train_indices)

Randomize Treatment Assignment

set.seed(123)
subjects <- 1:50

# Assign 25 to treatment, 25 to control
treatment <- sample(subjects, size = 25)
control <- setdiff(subjects, treatment)

Select Random Files

# Randomly select 5 CSV files from a directory
set.seed(101)
files <- list.files("data/", pattern = "\\.csv$")
selected <- sample(files, size = min(5, length(files)))

Shuffle a Deck of Cards

# Create and shuffle a deck
suits <- c("hearts", "diamonds", "clubs", "spades")
ranks <- c("A", 2:10, "J", "Q", "K")

deck <- expand.grid(rank = ranks, suit = suits)
deck <- deck[sample(nrow(deck)), ]  # Shuffle the deck

Sampling vs With-Replacement

Compare the two approaches:

set.seed(42)

# Without replacement - unique values only
sample(1:10, size = 5, replace = FALSE)
# [1]  7  1  9  6 10

# With replacement - duplicates possible
sample(1:10, size = 5, replace = TRUE)
# [1]  6  1  1  9  6

Common Variations

Task	Code
Sample n elements	`sample(x, n)`
Shuffle all elements	`sample(x)`
Sample with probability weights	`sample(x, n, prob = weights)`
Sample from custom pool	`sample(c("a", "b", "c"), 2)`

Using Probabilities

You can weight elements by probability:

set.seed(55)

# 10 elements with increasing probabilities
x <- 1:10
probs <- x / sum(x)  # probability proportional to value

sample(x, size = 5, prob = probs)
# [1]  7 10  9  5  8
# Higher values are more likely to be chosen